Posted by: Nerfherders
- [501035289] Mon, Mar 31, 2008, 16:46
I have the first draft ready to go, so I'll explain here the theory behind it before presenting data.I've used other ranking systems in the past but could never see the math behind them. How do you compare Eric Bedard to Mark Teixeria? Joe Nathan to Ichiro Suzuki? Alot of us use these ranking systems or use our own based on hunches or how a player 'feels' going into the year. The following is an attempt to quantify a player's direct point value to a rotisserie team based on past team's performances in the same or similar league. I can't say it will be particularly insightful in attempting to predict future performance if you choose to use anything other than past performance to do so (most of us would not). But with good projections, it will tell you how all those numbers will shake out for your team for overall points or in any particular category.Here's the basic principle:We look at stats for the same league, or similar leagues using the same rules, and re-sort those finishing stats so for each category we have it sorted by best to worst. We average over the leagues for each placement to come up with an average for that place. So that for example the average stat for 1st place in Runs is X, 2nd place in Runs is Y, etc etc. The following is the average finishing place for each stat for RIBC leagues, averaging twelve leagues over 4 years.[
R
RBI
SB
OBP
SLG
W
SV
K
ERA
WHIP
1
1086
1047
173
0.363
0.479
96
111
1149
3.57
1.23
2
1049
1013
160
0.358
0.469
93
101
1138
3.66
1.25
3
1037
1006
152
0.357
0.466
90
101
1107
3.79
1.27
4
1016
971
143
0.356
0.464
89
85
1094
3.82
1.28
5
1016
964
139
0.353
0.460
88
82
1071
3.92
1.29
6
996
955
134
0.353
0.458
86
81
1048
3.98
1.30
7
992
937
128
0.351
0.456
84
73
1051
4.03
1.31
8
981
924
126
0.350
0.453
83
71
1044
4.06
1.31
9
965
915
119
0.349
0.452
80
70
1021
4.13
1.32
10
957
902
110
0.347
0.450
79
69
1010
4.12
1.33
11
940
891
111
0.347
0.447
79
57
985
4.14
1.34
12
937
885
104
0.346
0.445
77
49
957
4.26
1.35
13
924
868
104
0.343
0.442
73
49
950
4.32
1.36
14
913
856
97
0.342
0.439
72
45
918
4.43
1.37
15
870
843
87
0.338
0.434
71
31
898
4.45
1.38
16
851
800
78
0.334
0.428
64
26
813
4.60
1.41
[The next step in the process is to calculate the regression of each category - how much of each category will represent a drop or rise by one placement.The next idea is coming up with a way to evaluate a player fairly regardless of playing time. To do this, I came up with what I call 'Stationary Level'. Stationary Level is the average 9th placement in RIBC. I chose 9th because it is the best placement in the league in which you gain nothing - you cannot move to AAA from AA, nor can you stay in RIBC. It is perfectly mediocre. The basic principle for the system is identifying player X's contribution to your team if you had player X and the rest of the team was Stationary Level. At this point I had to make some assumptions. I have to assume a team will use all of its innings. I have to assume that a team WILL NOT use all of its hitters games. I estimated the maximum to be about 1800. For saves I decided to use Pitchers Games instead of IP, so I came up with an estimate for that. I had to estimate team plate appearances and at-bats. For a full season, these are my estimates:IP: 1350Pitcher G: 441Hitter G: 1800Hitter PA: 7416Hitter AB: 6714Once category contributions have been calculated, that number is divided by the slopes we created earlier, and we have an actual point value contribution for each category.Now to the rankings - Top 20 hitters and top 20 pitchers from 2007:[
Rank
Full Name
Team
League
Pos
G
AB
Runs
Rbi
SB
OBP
Slug%
Wlks
Rv
RBIv
SBv
OBPv
SLGv
Tvalue
1
Rodriguez, Alex
NYY
AL
3B
158
583
143
156
24
0.422
0.645
95
4.42
5.42
2.47
4.22
5.98
22.51
2
Ramirez, Hanley
FLA
NL
SS
154
639
125
81
51
0.386
0.562
52
3.22
0.20
7.43
2.19
3.73
16.76
3
Holliday, Matt
COL
NL
LF
158
636
120
137
11
0.405
0.607
63
2.67
4.06
0.10
3.34
5.23
15.42
4
Wright, David
NYM
NL
3B
160
604
113
107
34
0.416
0.546
94
2.06
1.84
4.27
3.99
3.01
15.17
5
Ordonez, Magglio
DET
AL
RF
157
595
117
139
4
0.434
0.595
76
2.49
4.24
-1.16
4.86
4.52
14.95
6
Ortiz, David
BOS
AL
DH
149
549
116
117
3
0.445
0.621
111
2.74
2.96
-1.24
5.40
4.93
14.78
7
Rollins, Jimmy
PHI
NL
SS
162
716
139
94
41
0.344
0.531
49
3.95
0.84
5.52
-0.31
3.00
12.99
8
Jones, Chipper
ATL
NL
3B
134
513
108
102
5
0.425
0.604
82
2.74
2.43
-0.70
3.86
4.14
12.47
9
Reyes, Jose
NYM
NL
SS
160
681
119
57
78
0.354
0.421
77
2.52
-1.74
12.27
0.34
-1.14
12.25
10
Braun, Ryan
MIL
NL
3B
113
451
91
97
15
0.37
0.634
29
2.30
2.84
1.37
0.87
4.36
11.74
11
Utley, Chase
PHI
NL
2B
132
530
104
103
9
0.41
0.566
50
2.52
2.57
0.05
3.02
3.21
11.37
12
Pena, Carlos
TB
AL
1B
148
490
99
121
1
0.411
0.627
103
1.49
3.28
-1.60
3.14
4.55
10.87
13
Fielder, Prince
MIL
NL
1B
158
573
109
119
2
0.395
0.618
90
1.84
2.77
-1.53
2.61
5.05
10.74
14
Howard, Ryan
PHI
NL
1B
144
529
94
136
1
0.392
0.584
107
1.27
4.50
-1.55
2.34
3.71
10.27
15
Crawford, Carl
TB
AL
LF
143
584
93
80
50
0.355
0.466
32
1.24
0.53
7.38
0.33
0.42
9.90
16
Pujols, Albert
STL
NL
1B
158
565
99
103
2
0.429
0.568
99
1.08
1.63
-1.53
4.53
3.48
9.18
17
Granderson, Curtis
DET
AL
CF
158
612
122
74
26
0.361
0.552
52
2.83
-0.45
2.83
0.69
3.25
9.14
18
Byrnes, Eric
ARI
NL
LF
160
626
103
83
50
0.353
0.46
57
1.31
0.12
7.18
0.25
0.26
9.11
19
Sizemore, Grady
CLE
AL
CF
162
628
118
78
33
0.39
0.462
101
2.36
-0.31
4.06
2.56
0.32
8.99
20
Figgins, Chone
LAA
AL
3B
115
442
81
58
41
0.393
0.432
51
1.47
-0.03
6.08
1.85
-0.48
8.89
[
Rank
Full Name
Team
League
Pos
G
Inn
W
Sv
K
ERA
Whip
Wv
SVv
Kv
ERAv
WHIPv
TValue
1
Peavy, Jake
SD
NL
SP
34
223.3
19
0
240
2.54
1.06
3.14
-1.05
3.78
4.37
4.24
14.49
2
Putz, J.J.
SEA
AL
RP
68
71.7
6
40
82
1.38
0.7
0.95
5.71
1.48
2.43
3.23
13.80
3
Saito, Takashi
LAD
NL
RP
63
64.3
2
39
78
1.4
0.72
-1.00
5.66
1.56
2.16
2.81
11.19
4
Beckett, Josh
BOS
AL
SP
30
200.7
20
0
194
3.27
1.14
4.43
-0.93
2.24
2.12
2.65
10.52
5
Papelbon, Jonathan
BOS
AL
RP
59
58.3
1
37
84
1.85
0.77
-1.35
5.40
2.12
1.64
2.33
10.14
6
Bedard, Erik
SEA
AL
SP
28
182
13
0
221
3.16
1.09
1.20
-0.86
4.43
2.17
3.06
10.00
7
Santana, Johan
NYM
NL
SP
33
219
15
0
235
3.33
1.07
1.09
-1.02
3.69
2.15
4.00
9.91
8
Rodriguez, Francisco
LAA
AL
RP
64
67.3
5
40
90
2.81
1.25
0.55
5.83
2.08
1.09
0.35
9.90
9
Nathan, Joe
MIN
AL
RP
68
71.7
4
37
77
1.88
1.02
-0.14
5.12
1.21
1.99
1.57
9.75
10
Valverde, Jose
HOU
NL
RP
65
64.3
1
47
78
2.66
1.12
-1.55
7.16
1.56
1.16
0.94
9.28
11
Sabathia, C.C.
CLE
AL
SP
34
241
19
0
209
3.21
1.14
2.57
-1.05
1.42
2.73
3.18
8.85
12
Jenks, Bobby
CWS
AL
RP
66
65
3
40
56
2.77
0.89
-0.47
5.77
0.36
1.09
2.04
8.78
13
Cordero, Francisco
CIN
NL
RP
66
63.3
0
44
86
2.98
1.11
-2.07
6.55
2.03
0.90
0.97
8.38
14
Hoffman, Trevor
SD
NL
RP
61
57.3
4
42
44
2.98
1.12
0.33
6.31
0.04
0.81
0.84
8.32
15
Hamels, Cole
PHI
NL
SP
28
183.3
15
0
177
3.39
1.12
2.25
-0.86
2.04
1.67
2.69
7.78
16
Lackey, John
LAA
AL
SP
33
224
19
0
179
3.01
1.21
3.12
-1.02
0.51
3.09
1.82
7.52
17
Webb, Brandon
ARI
NL
SP
34
236.3
18
0
194
3.01
1.19
2.17
-1.05
0.81
3.26
2.27
7.46
18
Wagner, Billy
NYM
NL
RP
66
68.3
2
34
80
2.63
1.13
-1.13
4.60
1.51
1.26
0.95
7.19
19
Isringhausen, Jason
STL
NL
RP
63
65.3
4
32
54
2.48
1.07
0.06
4.30
0.25
1.33
1.19
7.13
20
Smoltz, John
ATL
NL
SP
32
205.7
14
0
197
3.11
1.18
0.97
-0.99
2.20
2.58
2.12
6.89
[After looking at the lists, the hitting seems to be pretty dead on. Pitching on the other hand is not what I would have expected. Saves are highly valued, probably moreso than anyone would like to admit. But one would have to admit that adding 40 saves to a roster would give ALOT of points to a team.Please feel free to add any comments, good or bad. edit: I had this all nicely in paragraphs, but using the tables removed them. I apologize for the lously formatting.
1
Seattle Zen
ID: 49112418 Mon, Mar 31, 2008, 17:44
I'm very glad you took a crack at this. One of my pet peaves in the fantasy baseball media is their failure to come up with an accurate way to measure past performance in a way you are attempting.
One thing I notices in the first chart is a few of the categories the numbers go up as you go down the column. For instance, strike outs: 6th place 1048. 7th place 1051. Stolen bases: 10th place 110, 11th place 111.
For the starting pitchers, there were multiple different negative figures for having no saves. Shouldn't every starter get equally punished for having no saves?
I understand what you are trying to do, but I don't think I completely understand what you actually did to get there. Could you explain in more detail? I'm far from a math wiz, but I can follow someone explaining just about anything.
2
Nerfherders
ID: 501035289 Mon, Mar 31, 2008, 19:00
I hadnt actually noticed the numbers going up like that. But I had simply averaged the nth position stat for each category across all leagues and that is how they came out. I probably should sort it descending after that - it probably wouldnt affect the outcome one way or another.
The differing negative values for saves is because I based saves on games played. The fewer the games played with 0 saves, the higher the number will be. It especially hurts middle relievers.
Okay so I'll take you through how I came up with A-Rod's run factor as an example.
You start with A-Rod's 158 games played. You subtract this from 1800 total games played by a typical RIBC team (an educated guess.) Youre assuming that the other 1642 games played by your team is at the Stationary Level (9th place).
There are factors for each category. The Stationary Level stat divided by games played will give you a factor for runs per game at the Stationary Level. You multiply this factor by the 1642 games. (.53611*1642=880.29) You add A-Rod's 143 runs scored to this total (1023.29). This is the number of runs you would score if you had A-Rod and 11 Stationary Level players.
If you look at the first chart you can see that the 1023 runs would take you from 9th to 3rd place, but based on the regression of the entire series it would actually be a little lower than this. To get this value you subtract the Stationary Level runs from A-rod+ (58.29). You then divide by the regression factor which happens to be about 13. You then get 4.42 - The expected points increase you would get for Runs by having A-Rod and 11 Stationary Level players, over 12 Stationary Level players.
3
biliruben
ID: 33258140 Mon, Mar 31, 2008, 19:11
Keep tweaking it until Johan is 1st and Putz is something other than 2nd. Then I'll believe it enough to bother with the nuts and bolts.
4
Seattle Zen
ID: 49112418 Mon, Mar 31, 2008, 19:38
Keep tweaking it until Johan is 1st and Putz is something other than 2nd.
You must have been asleep last year as Peavy was considerably better than Johan.
The differing negative values for saves is because I based saves on games played. The fewer the games played with 0 saves, the higher the number will be. It especially hurts middle relievers.
Sure, middle relievers should be punished for not getting saves, but starters never get saves and every starter should be punished equally in my mind.
5
biliruben
ID: 33258140 Mon, Mar 31, 2008, 20:08
Well, simply using one year of stats is a significant flaw in itself.
Back when I had time for such things, I took the average of the last 3 years to develop my prediction models. I then culled the universe of players down to those likely to be drafted in roto, standardized each cat (subtracted the mean/ divided by the SD).
Then bootstrapped 10,000 seasons and did took the average. Kidding.
My major unresolved issue was how to come up with a reasonable comparison of pitchers to hitters.
6
Khahan
ID: 486552412 Mon, Mar 31, 2008, 20:14
Sure, middle relievers should be punished for not getting saves, but starters never get saves and every starter should be punished equally in my mind.
I disagree here. I see how different starters should have different negative values for a stat such as saves. His charts are based on games played and there is a finite amount of games played for a team. If Johan Santana starts 32 games (games played=32) and Pedro Martinez starts 20 games (games played=20) then Pedro has depleted less resources that could turn into saves than Santana.
It may be a minor factor and you may not actually care one way or the other, but it is mathematically true. In this instance Pedro hurt you less in Saves because he used less games played leaving more open for somebody who could get a save.
This is just one minor aspect of the whole and I think the rest of the categories balance out fairly well.
Hope that makes sense. It does in my mind at least.
7
Seattle Zen
ID: 29241823 Tue, Apr 01, 2008, 01:33
Khahan: Pedro has depleted less resources that could turn into saves than Santana.
I disagree right back. Now that I think about it, I don't think starters should be punished at all for failing to get any saves.
I'm making the following assumptions. First, there are starting pitching slots on each fantasy roster. Starters are designated as such, relievers are relievers.
When Pedro pitches only 20 games, if you replace him, it has to be with another starter. A starter is not taking up space that could be used by a reliever who may get a save.
Yes, most fantasy baseball rosters have generic pitcher categories as well as starting and relievers and this over-complicates things.
Biliruben
Back when I had time for such things, I took the average of the last 3 years to develop my prediction models.
He's not predicting, he is measuring what each player was worth last year and I think he is onto something here. And, yes, Jake was better than Johan last year. He did use RIBC data from the past four years to determine the average numbers for each of the 16 places.
Nerf
I'm looking at the results and I'm seeing some really weird stuff. For instance, Jake Peavy, CC Sabathia and John Lackey all won 19 games, yet the Win Values were 3.14, 2.57, and 3.12. Why is there any difference? Both Jake and CC started 34 games, John 33. Beckett wins one more game and his value is a whopping 4.43. Why such a jump? If it is because he started only 30 games, I think that is a flaw in your system.
Eric Bedard won 13 games and got a Win Value of 1.20 while Santana wins 15 and his is only 1.05. Games played?
JJ Putz has a Win value of .95. John Smoltz is .97. John won 14 games, JJ won 6. Something's wrong there.
When I look at the top chart, "wins" is the tightest category. From 4th place to 12th, each additional win just about moves you up a level, unlike saves where there is huge disparity as you fall. I think you are overvaluing saves and undervaluing wins as well as incorrectly assigning value to wins. Furthermore, I don't think starters should be punished for not getting any saves as I argued above.
Saves are the only pitching stat based on games played. Everything else is based on IP and the idea is that you have a finite number of IP for your team so if you can get 20 wins out of 200 IP, that is better than 19 out of 240 IP. If you can get 6 wins out of 72 IP, that is about the same as getting 14 wins out of 205 IP. Its about efficiency, because youre assuming that the rest of your pitchers are Stationary Level. They will get you a win about every 17 IP.
Lets take an extreme example. Youre entire pitching staff is Stationary Level for 1349 IP, getting 80 wins. You sign a FA reliever and he pitches one inning and gets a win and gets hurt for the rest of the season. This gets you to your 1350 IP, but with the extra win you bump up to 81 wins. For wins, that one inning from that reliever is positive value. To be more precise, he gives you a 0.51 point value for wins.
In this example, the reliever with 1 IP and one win is slightly more valuable to wins than a starter with 210 IP and 13 wins.
9
Seattle Zen
ID: 49112418 Tue, Apr 01, 2008, 12:14
Nerf
In this example, the reliever with 1 IP and one win is slightly more valuable to wins than a starter with 210 IP and 13 wins.
I don't know if I would agree with that. See my bottom paragraph.
Everything else is based on IP and the idea is that you have a finite number of IP for your team so if you can get 20 wins out of 200 IP, that is better than 19 out of 240 IP.
But how much better?
Let's look at Josh Beckett and CC Sabathia. They have identical WHIP at 1.14. CC pitched about 40 innings more than Josh. So their WHIP value is 3.18 for CC, 2.65 for Josh. CC gets about a half a point for the extra innings. ERA for CC was 3.21 and Josh was 3.27, both outstanding ERA's. Again, because CC pitched about 40 more innings, his ERA value was 2.73 to Josh's 2.12.
Because Josh got one more win than CC in 40 less innings, his Win value was 4.43 while CC's was 2.57. That singular extra win gave Beckett an additional 1.86 points. That one extra win was nearly equal to pitching 200 innings at a 3.27 ERA. Come on, you've got to admit that this ain't right.
Its about efficiency, because you're assuming that the rest of your pitchers are Stationary Level.
I've been thinking about this. The Baseball Prospectus people have a value for players called VORP, Value over Replacement Player, that attempts to quantify how much better Player X is compared to someone with whom his team could replace him. If I remember correctly, they assume the replacement is what they call a "AAAA" level player, someone who is a little better than a AAA player, but not quite good enough to stick in the majors for good. You are assuming, on the other hand, that each player is taking innings away from a pitcher at Stationary Level. However, if your pitcher goes down, you have to replace him with someone from the waiver wire, not from Stationary Level.
Something to think about.
10
Khahan
ID: 486552412 Tue, Apr 01, 2008, 12:56
That singular extra win gave Beckett an additional 1.86 points.
You cannot make this statement to show the flaw in Nerf's system because there is a flaw in this statement. As it relates to Nerf's system, Beckett did not get 1 extra win that caused a 1.86 point swing. He got 1 extra win in 40 less innings which caused the point swing.
By chalking it all up to just 1 extra win you are only using half of the data and as such have half only half of the argument. Now, I'm not saying whether Nerf is right or wrong to do it this way. (I'm still reserving judgement on that), just that he IS doing it this way so when you evaluate the process that is what you have to take into account. You cannot simply ignore half of the values and expect to get a proper correlation.
However, if your pitcher goes down, you have to replace him with someone from the waiver wire, not from Stationary Level.
SZ, none of the points you raised show that you understand how Nerf is going about doing what he is doing. His baseline for stats does not actually come from what a player can do. His baseline from stats comes from the grand total a roto team needs to accumulate to win a category. His baseline is what the 9th place TEAM accumulates. Not what a certain player accumulates. He is then using this information to show how any given individual player contributed to the team goals.
Whether that player was the first off the board in the draft or a WW pick up at the allstar break is irrelevant. What matters is 'what did that player contribute towards the teams goal and how many resources did he use to contribute.'
Simply put a player with 1 RBI in 1 game played was a much more efficient contributor towards the 1047 RBi's needed to win the category than somebody who contributed 1 rbi in 10 games played (because his contribution to the 1047 vs resources used ratio was much better).
For pitching you are not just comparing SP to SP. You are first comparing the contributions of any 1 starting pitcher to the 96 W, 111 Saves, 1149 K's, 3.57 ERA and 1.23 whip needed to win those categories. Its only after you get those values that you can then try and compare 1 pitcher to another. And you do so by comparing those contributions vs resources used.
Maybe I'm misinterpreting what you are saying but every argument you are making indicates you are skipping over the first step.
I have to agree with Khahan here. In the comparison of Beckett's numbers to CC's numbers, you are starting with the assumption that all of the other innings are Stationary Level. So in order to fill out the innings that CC contributed, in those 40 extra innings the Stationary pitchers are going to give you 2-3 extra wins. The regression of wins is 1.82, so the difference of 23 wins to 19 in the same number of IP is worth 2 points. Wins is a tightly packed stat from top to bottom so it doesnt take much to move up or down.
12
Seattle Zen
ID: 49112418 Tue, Apr 01, 2008, 13:20
He got 1 extra win in 40 less innings which caused the point swing.
If you bothered to read my whole post, I REPEATEDLY mentioned that CC pitched 40 innings more than Beckett. Yes, Josh Beckett got 1 more win in 40 FEWER INNINGS, and he earned 1.86 MORE points for it. He also earned 2.12 points for throwing 200 innings of 3.27 ERA. I think there is SERIOUS problems with a system that rewards someone for one more win in 40 fewer innings nearly as much as his outstanding 3.27 ERA. What do you have to say about that? Nothing, because you are "reserving judgment" on that. Sounds like YOU are the guy who is not understanding what Nerf is doing.
If you are too timid to make a judgment upon Nerf's system, don't make a fool out of yourself criticizing others who are trying to help.
13
dpr
ID: 1733917 Tue, Apr 01, 2008, 13:35
looking at his data if you have all stationary level player his ERA would be 4.13. Simplifying to say beckett pitched 1/7 of the innings with the rest going to stationary players the team ERA moves to 4.01. This would move you up roughly 2 points based on the average results.
For wins you obviously get becketts 1 additional win and then you also get 40 innings of stationary innings to amass more wins. based on averages nerf says this is 2-3 wins. So having beckett instead of CC would give a team 3-4 more wins which again moves u up roughly 2 points in the standings.
In the context of what he is doing his results make sense to me
14
Khahan
ID: 46121614 Tue, Apr 01, 2008, 16:49
I REPEATEDLY mentioned that CC pitched 40 innings more than Beckett
Yes you did acknowledge the difference but not when you discussed results: That singular extra win gave Beckett an additional 1.86 points
Right before this you mention the 40 innings but you leave it out when you want to make your point.
I think there is SERIOUS problems with a system that rewards someone for one more win in 40 fewer innings nearly as much as his outstanding 3.27 ERA
Who is being rewarded here? These resulst are showing us that 1 more win in 40 less innings is efficient. I think my 7 year old could have told all of us that. What this system does is tell us how much more efficient.
There is no reward or punishment. The only people getting anything are us (apparently some of us are getting more than others). And all we get is a look at how efficiently each player contributed to different categories. Nothing more.
I am still trying to get my head around the saves issue. When I ran the numbers for my keeper league (10 team, 1600 IP limit) the negatives for saves for starters come out to about -0.40 or so. For this league the numbers have a much greater dispersion, and there are some worthy SP in the rankings who really get hurt with that -1.
So for those who have seen these rankings, does it look like something that could be useful to you?
16
Guru
ID: 330592710 Wed, Apr 02, 2008, 17:24
A couple of observations:
I'm not sure why you chose GP to "scale" the save category. It seems like IP should be the basis for all pitching cats, since that is what is limited in this type of league. We can only use 1350 IP. But it doesn't matter how many games we get.
Having said that, if everything is scaled based on IP, then the rankings would probably tend to favor relievers. But if a team "needs" (or at least plans) to amass 1350 IP, it can't do that with mostly relievers.
It would be interesting to use this analysis to develop some regression factors that would relate to each category. The factors could then be applied to projected stats to develop relative player rankings for a new season.
Ive been working on it for the last little bit and I think I have come up with something to normalize saves using IP.
The idea is to split SP innings and RP innings - SP's get an allotment of 1000 innings, and RP an allotment of 350 innings.
The rate of SP innings for saves is zero. Which means that for SP's, their saves factor is zero. RP's saves rate is a hair under .20 per inning to get to the Stationary 69.75 saves. Now this of course is making a major assumption that all teams have 5 relievers playing full time. The other problem with this is that we are forced to categorize pitchers as one or the other, which at any given time they are, but can switch in the middle of a season.
But the numbers are looking real good. The saves cat for a team is adding up to the proper +/- now. Meaning my projections for my pitchers are 80 saves, which should be about a 3 point increase, but with all the -1's in the SP slots, it was coming out to about 0. Now its coming out to three.
18
holt
ID: 341542412 Sun, Apr 06, 2008, 02:37
Someone may have already pointed this out, but you can't determine what player A is worth compared to player B without doing it within the context of a complete roster. Mariano Rivera may be worth 3, 5 or, or 10 roto points for your team depending on the number of other closers you have on your roster (just throwing numbers out there). A player like Pierre's value will vary wildly depending on the rest of your roster. Well, not only your roster, but also the rosters of the 15 other teams.
If you wanted to take the time, you could look at all the ribc leagues from last season and break down individual player values for each team in a pretty accurate way, but the individual player values would vary from league to league.
If you wanted to develop a system like this so that you can draft a team in a more scientific way, you'd have to use a draft tracker system that constantly updates values based on every teams complete roster (that's nothing new). The major problem is that you'd just be dealing with projections, so it would still be more of an art than a science in my opinion.
Anyway, point is that you can't determine an absolute value for a player within a specific roto system. That's what separates roto from points systems. Every RIBC players value fluctuates based not only on his own team's roster but also on the roster makeups of the other 15 teams in the league.
btw, has anyone ever spent much time comparing pre-season projections to actual end-of-year stats? You have to wonder what % of the players actually perform with 10-15% of their projections. I'm probably like a lot other roto players in that I use projections as a rough starting point for sorting all the players, but then when it comes to narrowing down players for an actual draft pick I research stats and news and come up with my own "projections". Projections are never "money in the bank" in any case. Far from it.
19
Khahan
ID: 486552412 Sun, Apr 06, 2008, 09:10
Holt, Nerf is not 'determining' a value for a player. He's measuring a contribution. What value you pull from that is up to you.
Also, he is not determining the value of 1 player vs another. He is determining the contribution of a player to an overall goal. That number can be used to make a comparison. But that does not seem to be the objective to me.
For what Nerf has set out to do (measure the contribution of a player last year to an overall goal of X in category Y) his system is pretty good except for the pitchers: Innings pitched vs games started. This seems completely arbitrary. I don't understand why he didn't simply use a total innings pitched based on the average # of innings allowed in each league.
This seems to be the only part of the system where you made a random adjustment to get results that looked better to you. A rookie who throws 1 inning and gets 3 k's and 1 save is going to be off the charts here. But considering that this is nothing more than a tool that should be utilized with other research, I don't particularly see a problem with that. It will be looking at the Yahoo rankings at the allstar break and searching based on BA. There are going to be about 15 guys who are 1 for 1 or 2 for 2 and filter as the best BA. But we all know to avoid them for other reasons.
Here is a top 20 rankings based on saves as an innings allotment, where starters get a zero value for saves and relievers get a value based on an allotment of 350 IP.
Rank
Full Name
Team
League
Pos
G
Inn
W
Sv
K
ERA
Whip
Wv
SVv
Kv
ERAv
WHIPv
TValue
1
Peavy, Jake
SD
NL
SP
34
223.3
19
0
240
2.54
1.06
3.14
0.00
3.78
4.37
4.24
15.54
2
Putz, J.J.
SEA
AL
RP
68
71.7
6
40
82
1.38
0.7
0.95
5.02
1.48
2.43
3.23
13.11
4
Beckett, Josh
BOS
AL
SP
30
200.7
20
0
194
3.27
1.14
4.43
0.00
2.24
2.12
2.65
11.45
7
Santana, Johan
NYM
NL
SP
33
219
15
0
235
3.33
1.07
1.09
0.00
3.69
2.15
4.00
10.93
6
Bedard, Erik
SEA
AL
SP
28
182
13
0
221
3.16
1.09
1.20
0.00
4.43
2.17
3.06
10.86
3
Saito, Takashi
LAD
NL
RP
63
64.3
2
39
78
1.4
0.72
-1.00
5.11
1.56
2.16
2.81
10.64
11
Sabathia, C.C.
CLE
AL
SP
34
241
19
0
209
3.21
1.14
2.57
0.00
1.42
2.73
3.18
9.90
5
Papelbon, Jonathan
BOS
AL
RP
59
58.3
1
37
84
1.85
0.77
-1.35
4.95
2.12
1.64
2.33
9.69
8
Rodriguez, Francisco
LAA
AL
RP
64
67.3
5
40
90
2.81
1.25
0.55
5.19
2.08
1.09
0.35
9.26
9
Nathan, Joe
MIN
AL
RP
68
71.7
4
37
77
1.88
1.02
-0.14
4.43
1.21
1.99
1.57
9.06
10
Valverde, Jose
HOU
NL
RP
65
64.3
1
47
78
2.66
1.12
-1.55
6.67
1.56
1.16
0.94
8.79
15
Hamels, Cole
PHI
NL
SP
28
183.3
15
0
177
3.39
1.12
2.25
0.00
2.04
1.67
2.69
8.65
16
Lackey, John
LAA
AL
SP
33
224
19
0
179
3.01
1.21
3.12
0.00
0.51
3.09
1.82
8.54
17
Webb, Brandon
ARI
NL
SP
34
236.3
18
0
194
3.01
1.19
2.17
0.00
0.81
3.26
2.27
8.51
12
Jenks, Bobby
CWS
AL
RP
66
65
3
40
56
2.77
0.89
-0.47
5.28
0.36
1.09
2.04
8.29
14
Hoffman, Trevor
SD
NL
RP
61
57.3
4
42
44
2.98
1.12
0.33
5.97
0.04
0.81
0.84
7.98
13
Cordero, Francisco
CIN
NL
RP
66
63.3
0
44
86
2.98
1.11
-2.07
6.12
2.03
0.90
0.97
7.95
20
Smoltz, John
ATL
NL
SP
32
205.7
14
0
197
3.11
1.18
0.97
0.00
2.20
2.58
2.12
7.88
23
Harang, Aaron
CIN
NL
SP
34
231.7
16
0
218
3.73
1.14
1.22
0.00
2.27
1.13
3.06
7.69
22
Vazquez, Javier
CWS
AL
SP
32
216.7
15
0
213
3.74
1.14
1.16
0.00
2.61
1.03
2.86
7.67
21
Boldwin
ID: 463471413 Mon, Apr 14, 2008, 21:30
Judging what fudge factor each league assigns closers, ie. when their closer runs begin is unique to each league and we all know they go later than their true value in points. Are they even undervalued in auction style? Prolly even there. I think leaving them at their true contribution level in these rankings is a useful tool to counterbalance that drafting fact of life because for trade value purposes they come closer to their true value and assessing your roster's true value also benefits.
Rate this thread:
If you wish, you may rate this thread on scale of 1-5. Ratings should indicate how valuable or interesting you believe this thread would be to other users of this forum. A '5' means that this thread is a 'must read'. A '1' means that this is a complete waste of time.
If you have previously rated this thread, rating it again will delete your previous rating.
If you do not want to rate this thread, but want to see how others have rated it, then click the button without entering a rating, or else click here.